diff --git a/CHANGELOG.md b/CHANGELOG.md index edb6de7..d2fe48e 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -7,6 +7,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0 ## [Unreleased] +## [0.5.2] - 2026-10-18 + ### Added - **Deep links for your notes (`vixel://open?…`).** Right-click any item (or use the inspector) → Copy Link / Copy Markdown Link, paste into Obsidian / Notes / anywhere; @@ -36,6 +38,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0 points you to System Settings → Keyboard → Dictation instead of downloading anything itself. Windows is not covered yet (no offline engine met the quality bar). +### Fixed +- **Text in images: long lines lost most of their characters** (#3). Each recognized line + was squeezed into a 320 px wide input, so a 30–40 character line (screenshots, articles) + came back with large gaps. Lines now get a width that matches their aspect ratio, as in + PaddleOCR itself; the reporter's screenshot now reads line for line like their local + PaddleOCR. Existing libraries re-run OCR automatically after updating. +- The player's segment strip said "片段 2" for a two-segment video, which read like + "segment #2"; it now says "2 个片段". + ### Decided against (for now) - **Search by a sound clip.** Measured on the demo footage: audio queries against the joint frames+audio segment vectors rank the right video first only 3/12 times (the picture diff --git a/electron-builder.yml b/electron-builder.yml index 49cfb8c..724ae78 100644 --- a/electron-builder.yml +++ b/electron-builder.yml @@ -20,6 +20,8 @@ files: # 构建产物 / 模型 / 原生运行时都不进 asar:模型和 LiteRT 走 extraResources,dist 是 CLI,release 是上一次的打包输出 # 注意:不要在 mac: / win: 下写 files —— 平台级 files 会整个替换这里的列表,上面的排除全部失效 - '!{release,dist,resources}/**' + # 本地工作目录:worktree(含整份源码和资源副本)、server 的 Node ABI 依赖、自定义网页构建、测试输出 + - '!{.claude,.git,server,e2e-out,coverage}/**' # onnxruntime-node(人脸 / OCR)自带全部平台的二进制,只留目标平台的 - '!node_modules/onnxruntime-node/bin/napi-v*/!(${platform})/**' asar: true diff --git a/package-lock.json b/package-lock.json index 061fb46..f4735b6 100644 --- a/package-lock.json +++ b/package-lock.json @@ -1,12 +1,12 @@ { "name": "vixel", - "version": "0.5.1", + "version": "0.5.2", "lockfileVersion": 3, "requires": true, "packages": { "": { "name": "vixel", - "version": "0.5.1", + "version": "0.5.2", "hasInstallScript": true, "license": "MIT", "dependencies": { diff --git a/package.json b/package.json index 50c5e72..a669da9 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "vixel", - "version": "0.5.1", + "version": "0.5.2", "description": "Private, offline AI search for your photos, videos and audio", "main": "./out/main/index.js", "author": "Vixel Team",